$ Web Access for Agents

Works on sites
others can't.

Give your AI agent live web access to any public site. 15 years of unblocking infrastructure behind a free MCP server. 5,000 requests per month, no credit card.

5,000
free requests/month
8+
framework integrations
< 4s
avg response time
99.7%
success rate
195
countries

Trusted by 20,000+ customers and 70% of AI labs

View pricing

Deloitte
McDonald's
Moody's
NBC Universal
Nokia
Oxford
Pfizer
Shopee
Taboola
eToro
United Nations
Club Med
SOC 2 Type II
ISO 27001
GDPR
CCPA
CSA STAR
View Trust Center
$ Integration

Two minutes to integrate.

One MCP config. One npm command. One API call. Works with every major agent framework and IDE. Copy, paste, run.

claude_desktop_config.json
// claude_desktop_config.json
{
  "mcpServers": {
    "Bright Data": {
      "command": "npx",
      "args": ["@brightdata/mcp"],
      "env": {
        "API_TOKEN": "your-api-token"
      }
    }
  }
}

What your agent can do.

Your agent reads Wikipedia. Does it read Amazon, LinkedIn, Google Maps? With Bright Data web access, it reads anything.

Read any public page

Amazon product pages, Google results, LinkedIn profiles, government databases. If a human can see it in a browser, your agent can read it.

Search the web

Structured search results from Google, Bing, or Yandex. Full SERP data including snippets, links, and metadata. Filter by language, region, or date.

Extract structured data

Get JSON, not HTML. Product prices, review ratings, article text, contact information. Custom extraction schemas for any page structure.

Monitor changes

Track price changes, new listings, content updates. Set up recurring checks and get notified when data changes. Build reactive agents.

Try the browser in your agent.

Start free
$ Discovery

Agents that find
their own data.

Stop hardcoding URLs. Discovery tools let your agents autonomously locate the right data sources, by intent, in real time.

Discover API

Intent-rankedlive discovery

Your agent describes what it needs, the Discover API returns ranked domains and pages by relevance. Agents autonomously find the right data sources without hardcoded URL lists.

Natural language intent, agents describe their data need
Returns ranked, relevant URLs automatically
No hardcoded site lists, agents adapt dynamically
Feeds directly into scraping or extraction tools

Fast SERP

< 1sresponse time

Sub-second search results for time-sensitive agent workflows. When your agent needs to verify a fact, check a price, or find a URL. Fast SERP delivers the top 10 results instantly.

Under 1 second response time
Top 10 Google results, optimized for speed
Perfect for agent verification and fact-checking loops
Low latency for real-time conversational agents

See it work on any site.

Try the playground
$ Pricing

Start free, scale when ready.

No credit card for free tier. No surprise charges. Upgrade from your dashboard when you need more.

FREE
Prototype
$0
MCP server + search + scrape
5,000 requests/month
Full site coverage
No credit card required
Community support
Start free
PRO
Production
From $0.001/req
Everything in Free
60+ tools including browser automation
Web data APIs & e-commerce scrapers
Pay-as-you-go, scale when ready
Priority support
View pricing
ENTERPRISE
Full platform
Custom
Everything in Pro
Datasets + Real-Time + Agents
Dedicated account team
Custom SLAs & volume discounts
Compliance documentation & DPA
Contact sales

Bright Data's browser infrastructure helps scale our AI agents for complex tasks. It lets us focus on delivering real value to our customers instead of wrestling with browser infrastructure.

Devi Parikh
Co-Founder, Yutori
$ FAQ

Common questions

MCP (Model Context Protocol) lets AI agents call external tools. Bright Data's MCP server gives your agent the ability to search the web, scrape any page, and extract structured data, all through a single npm command. It works with Claude Desktop, Cursor, Windsurf, Copilot, and any MCP-compatible client.
Your requests simply stop working until the next month. No surprise charges, no billing without your explicit consent. When you are ready to scale, upgrade to Pro (pay-as-you-go) from your dashboard. Downgrade anytime.
We have native integrations with LangChain, CrewAI, AutoGen, LlamaIndex, OpenAI Agents, Semantic Kernel, and n8n. For any other framework, use the REST API directly, it works with any HTTP client.
Yes. We have residential IPs in 195 countries. Your agent can access content as if browsing from any location worldwide. Specify the target country in your API request.
5,000 requests per month, no per-minute rate limit. Requests are processed concurrently with automatic scaling. If you need higher throughput, upgrade to Pro for unlimited concurrency.
A simple HTTP request fails on most modern websites due to anti-bot defenses, CAPTCHAs, JavaScript rendering, and IP blocks. Bright Data handles all of this automatically, your agent sends one API call and gets clean, structured data back. No Playwright, no Puppeteer, no proxy management.
Yes. SOC 2 Type II certified, ISO 27001, GDPR and CCPA compliant. Enterprise plans include DPA agreements, custom compliance documentation, and dedicated security review. Won court cases against Meta and X for ethical data collection.

Your agent can read Wikipedia.<br/>Can it read Amazon?

Install in 2 minutes. 5,000 requests free. No credit card.

No credit card required for free tier